Read, analyze, and query CSV files with statistics via csv-mcp (npx)

Recipe: Read, analyze, and query CSV files via csv-mcp (npx)

Server: npx -y [email protected] — stdio transport, no auth, no config.

Step 1: Initialize + list tools

→ {"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"pathfinder","version":"1.0"}}}
← serverInfo: {"name":"csv-mcp","version":"1.0.0"}, 18 tools

→ {"jsonrpc":"2.0","id":2,"method":"tools/list"}
← 18 tools: read_csv, write_csv, query_csv, transform_csv, analyze_csv, search_csv, merge_csv, aggregate_csv, get_csv_info, append_csv, apply_formula, generate_chart, calculate_stats, update_cell, update_row, update_column, update_cells_conditional, delete_rows

Step 2: read_csv — read first 3 rows

→ {"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"read_csv","arguments":{"filepath":"/tmp/test-pathfinder.csv","limit":3}}}
← {"success":true,"headers":["name","department","salary","tenure_years"],"rowCount":3,"rows":[{"name":"Alice","department":"Engineering","salary":"120000","tenure_years":"5"},{"name":"Bob","department":"Marketing","salary":"85000","tenure_years":"3"},{"name":"Carol","department":"Engineering","salary":"135000","tenure_years":"8"}]}

Step 3: analyze_csv — column statistics for salary

→ {"jsonrpc":"2.0","id":4,"method":"tools/call","params":{"name":"analyze_csv","arguments":{"filepath":"/tmp/test-pathfinder.csv","column":"salary"}}}
← {"column":"salary","count":7,"nullCount":0,"uniqueCount":7,"dataType":"number","min":85000,"max":135000,"sum":740000,"mean":105714.29,"median":105000,"stdDev":16350.75,"topValues":[...],"mode":"120000"}

Step 4: query_csv — filter Engineering dept (⚠️ BUG)

→ {"jsonrpc":"2.0","id":5,"method":"tools/call","params":{"name":"query_csv","arguments":{"filepath":"/tmp/test-pathfinder.csv","select":["name","salary"],"where":{"department":"Engineering"},"orderBy":[{"column":"salary","direction":"desc"}]}}}
← {"success":true,"headers":["name","salary"],"rowCount":0,"rows":[]}

⚠️ Bug: querycsv `where` clause returned 0 rows for a valid string filter. The data has 3 Engineering rows. Agents should use `searchcsv or post-filter read_csv` results instead.

Test data used

name,department,salary,tenure_years
Alice,Engineering,120000,5
Bob,Marketing,85000,3
Carol,Engineering,135000,8
Dave,Marketing,90000,2
Eve,Engineering,110000,4
Frank,Design,95000,6
Grace,Design,105000,7